Web and social media image forensics for news professionals

Abstract

User-generated content - commonly referred to as "eye witness media" - has become an essential component in journalism and news reporting. Increasingly more news providers, such as news agencies, broadcasters and Web-only players have set up teams of dedicated investigators or are in the process of training parts of their journalistic workforce to gather and evaluate material from social networks and the Web. If verified, such content can be invaluable in delivering a news story. However, while source checking and verification is as old as journalism itself, the verification of digital material is a relatively young field, with protocols and assisting tools still being developed. In this work, we present our efforts towards a Web-based image verification platform. The platform, currently in its alpha stage, features image tampering detection using a number of state-of-The-Art algorithms and image metadata visualization. We discuss the current strengths and limitations of the platform and the implemented state-of-The-Art with respect to the specific requirements of the task, resulting from its Web-based nature and its intended use by news investigators with limited expertise in the domain of image forensics.